On Jul 2, 2012, at 7:37 AM, Kristian Fiskerstrand wrote: > This is rarely necessary and should be done only when the > default configuration selects the wrong mutex implementation. Is Gentoo3 or Gentoo4 selecting the wrong mutex implementation?
The deep underlying reason for using pthread_mutex is Do multi-thread and multi-process locks use the same mechanism? The pthread_mutex implementation using futexes is likely intrinsically faster, and pthread_mutexes unify interprocess and inter-thread locks in the same mechanism. Statically library linking in a multlib environment has some tricky build prerequsites to find pthread_mutex() in an arch-specific static library on a multlib peculiar path. Add the build prereqa, not change the mutex implementation, is the better solution imho. The --enable=o_direct should likely not be used on linux too: http://kerneltrap.org/node/7563 I've personally tried O_DIRECT with no improvement tuning bdb in rpm. But they are all build options to use however one wishes: but the criteria to decide what options to use are rather murky. Here is what I have used for years (including RHL/RHEL3/RHEL4 Berkeley DB) can be found in many dbXY.spec files is mostly similar to what you used: %configure -C --srcdir=../dist/ \ --includedir='${prefix}/include/%{name}' \ --program-transform-name=s,^db,%{name}, \ --enable-compat185 --enable-dump185 \ --enable-shared --enable-static \ --enable-cxx \ --enable-sql \ --enable-sql_codegen \ --enable-stl \ %ifarch %{java_arches} --enable-java \ %else --disable-java \ %endif # --enable-jdbc \ # --with-cryptography=yes \ # --with-stacksize=SIZE \ # --with-uniquename=SUFFIX \ # --with-pic \ # --with-gnu-ld \ # --enable-jdbc --with-jdbc=DIR \ # --enable-perfmon-statistics \ # --enable-dtrace \ # --enable-systemtap \ # --enable-umrw \ # --enable-test \ # --enable-debug --enable-debug_rop --enable-debug_wop \ # --enable-diagnostic \ DIsclaimer: They are all build options to use however one wishes: but the criteria to decide what options to use is rather murky. YMMV, everyone's does. hth 73 de Jeff
_______________________________________________ Sks-devel mailing list Sks-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/sks-devel